The very first and largest usable Burmese Language GPT in Myanmar
- Free to use and open-source
- Lightweight and Accurate
- Burmese + International Languages (Total 61 Languages)
- And, It is Awesome!
MyanmarGPT is the very first and largest usable Burmese language GPT in Myanmar with strong community contributions. It was created by me, Min Si Thu.
These two models are trained by using private property datasets, manually cleaned by Min Si Thu.
There are two versions of MyanmarGPT at the moment, 2023 December.
- MyanmarGPT - 128 M parameters
- MyanmarGPT-Big - 1.42 B parameters
Extended, released in 2024, January 28.
- MyanmarGPT-Chat - 128 M parameters
Released in 2024, February 23.
- MyanmarGPTX - Faster, Lightweight and Multiplatform
MyanmarGPT is 128 million parameters Burmese Language Model. It is very lightweight and easy to use on all devices.
MyanmarGPT-Big is a 1.42 billion parameters Multi-Language Model. It is an enterprise-level LLM for Burmese Language mainly and other languages. Currently supports 61 Languages.
Fine-tuned on MyanmarGPT, question answering model for the Burmese language. With the knowledge of "A Brief History of the World"
How to use - Tutorial on Building MyanmarGPT-Chat on local machine
Fine-tuned on MyanmarGPT-Chat, question answering model for the Burmese language. Faster, lightweight and multiplatform available model.
Install hugging face transformer
pip install transformers# Using Pipeline
from transformers import pipeline
pipe = pipeline("text-generation", model="jojo-ai-mst/MyanmarGPT")
outputs = pipe("အီတလီ",do_sample=False)
print(outputs)# Using AutoTokenizer and CausalLM
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("jojo-ai-mst/MyanmarGPT")
model = AutoModelForCausalLM.from_pretrained("jojo-ai-mst/MyanmarGPT")
input_ids = tokenizer.encode("ချစ်သား", return_tensors='pt')
output = model.generate(input_ids, max_length=50)
print(tokenizer.decode(output[0], skip_special_tokens=True))# Using Pipeline
from transformers import pipeline
pipe = pipeline("text-generation", model="jojo-ai-mst/MyanmarGPT-Big")
outputs = pipe("အီတလီ",do_sample=False)
print(outputs)# Using AutoTokenizer and CausalLM
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("jojo-ai-mst/MyanmarGPT-Big")
model = AutoModelForCausalLM.from_pretrained("jojo-ai-mst/MyanmarGPT-Big")
input_ids = tokenizer.encode("ချစ်သား", return_tensors='pt')
output = model.generate(input_ids, max_length=50)
print(tokenizer.decode(output[0], skip_special_tokens=True))[<iframe width="898" height="505" src="https://www.youtube.com/embed/RujWqJwmrLM" title="Chat GPT (AI) ကို မြန်မာလို သုံးစွဲနိုင်တော့မလား - DVB Youth Voice" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" allowfullscreen></iframe>](https://youtu.be/RujWqJwmrLM)
- Min Si Thu